The body plan, size, and behavior of predator and prey can all have important consequences for the success of predators and how they subdue and consume prey (Endler, 1991). For example, body plans with spines, teeth, or claws can impede the ability of predators to subjugate and consume prey (Crofts and Stankowich, 2021) and, in extreme cases, kill a predator (Kornilev et al., 2022). Even when a body plan provides little protection against predators, other attributes of prey can provide defenses such as chemical defenses (Eisner and Meinwald, 1966; Geffeney et al., 2002) and a wide variety of rapid escape behaviors with special neural circuitry such as the Mauthner-initiated C-starts of fishes (Eaton et al., 2001), the tail flips of crayfish (Edwards et al., 1999), and the withdrawal reflex of earthworms (Gunther, 1971). Increased prey size can also increase protection against predators in two major ways. First, increased size is correlated with the ability to generate greater forces that could more effectively deter a predator. Second, size by itself can also prevent predators from eating large prey, if like snakes, the predators do not take bites out of their prey (Price et al., 2015). Large disparities in sizes can even reverse which species is predator or prey as when crocodilians eat juvenile pythons and large pythons eat small crocodilians (Snow et al., 2007).

The more than 3,500 extant species of snakes are all predators (Pough et al., 2016) that have evolved specialized body plans and behaviors that allow them to swallow diverse prey whole (Cundall and Greene, 2000; Cundall and Irish, 2022). Compared to most vertebrates commonly eaten by snakes (Colston et al., 2010; Grundler and Rabosky, 2021), many arthropods, especially crustaceans, have body plans with an armor-like exoskeleton, spiny projections, and powerful claws that impose substantial challenges for predators. Nonetheless, specializing on crustacean prey has evolved in homalopsid snakes (Jayne et al., 2018) and independently in two genera (Liodytes and Regina) of natricine snakes (Gibbons and Dorcas, 2004). Some of these snakes lack obvious anatomical specializations, but they have evolved behavioral specializations for exploiting formidable crustacean prey (Jayne et al., 2018). Furthermore, some of these species can detect (Jackrel and Reinert, 2011) and exploit crustaceans when they are most vulnerable soon after molting (Jayne et al., 2002). Hence, these taxa are well suited for gaining insights into the primacy of behavior in determining the outcomes of predator-prey interactions, including some profound changes in the risks posed by prey with a given body plan.

After striking their prey, the predatory behaviors of different snake species may vary by: (1) immediately beginning to swallow live, moving prey, (2) rapidly injecting venom and then releasing prey, (3) holding the prey with the jaws during envenomation until the prey is immobilized or dead, or (4) using coils of the body to either restrain or kill prey (Cundall and Greene, 2000). Interestingly, the use of either venom or constriction to kill prey has evolved independently in many different lineages of snakes (Greene and Burghardt, 1978; Palci et al., 2021). This rampant convergent evolution could imply that killing prey before swallowing conveys significant benefits, perhaps including the capacity to consume larger prey. However, empirical data relevant to this plausible suggestion are quite scanty, and approaching this issue by comparing different species of snakes is often complicated by the different types and sizes of prey that are consumed by snake species that also often differ in size. Recently developed methods for quantifying the maximal gape of snakes and cross-sectional area of prey (Jayne et al., 2018, 2022; Gripshover and Jayne, 2021) can reduce some of these difficulties. However, despite the maximal gape of snakes imposing a clear upper limit for prey size, accounting for behavior is vital as snakes may choose prey that are only a small subset of what their gape permits.

In this study we focused mainly on the feeding behavior of Liodytes rigida for the following three reasons. First, when eating crayfish this species uses two of the key behaviors, coiling, and envenomation, that are often assumed to facilitate eating large or formidable prey. Second, a prior study (Tumlison and Roberts, 2018) suggested that the predatory behavior of this species is a highly stereotyped fixed action pattern. However, Tumlison and Roberts (2018) only used hard-shell crayfish without methodically manipulating and accounting for prey size, and both of these factors seem likely to affect the risk and difficulty of handling prey. Hence, an open question for this system and many others is whether apparent behavioral stereotypy mainly reflects animals being exposed to a very narrow range of relevant stimuli. Finally, the phylogeny of this species and its relatives is well resolved (Figueroa et al., 2016), and the feeding behavior of Liodytes alleni, which is the sister species of L. rigida, is well-studied (Gripshover and Jayne, 2021), both of which facilitate gaining insights into the evolution of possible specializations associated with this unusual diet.

We fed L. rigida, with known maximal gape, hard- and soft-shell crayfish with a large range in size to test the following two alternative hypotheses regarding predatory behaviors. First, L. rigida uses stereotyped feeding behaviors that are unaffected by crayfish size and hardness. Second, plasticity of behavior occurs with changes that are likely to reflect different risks and demands of handling prey. For example, if prey are very small, they are probably less dangerous, and coiling around them may be mechanically more difficult. Additionally, the decreased hardness and mobility of crayfish after molting would be expected to decrease the risk of injury, the difficulty of subduing prey, and the chance of escape, all of which could reduce the probability of both envenomation and coiling. For Liodytes pygaea, the sister species of the two crayfish-eating species of Liodytes, we offered mostly large vertebrate prey, to determine whether some of the behaviors that facilitate eating crayfish were likely to have been present before the transition from vertebrate to invertebrate prey. Finally, we also compared L. rigida to other crustacean specialists including L. alleni to gain further insights into the potential benefits of evolving specialized behaviors or a venom apparatus.

Our results emphasize how size and other attributes of prey can profoundly affect the behaviors used by a predator to capture and consume prey. For hard-shell crayfish, the occurrence of envenomation and swallowing prey from tail to head were indeed highly stereotyped for L. rigida as previously reported (Tumlison and Roberts, 2018). Nonetheless, many other aspects of predatory behavior varied substantially with variation in prey; hence, we evaluate whether any evidence suggests that decreased stereotypy in the predatory behavior of L. rigida is associated with attributes of prey that likely make them easier to subdue and consume. Two key issues are what stimuli elicit specialized behaviors and what, if any, benefits do those specialized behaviors convey. Additionally, we use variation within a species as well as comparative data to gain general insights into the evolution of specializations, or the lack thereof, for gape-limited predators that exploit unusual or formidable prey.

The variable use of envenomation and coiling by L. rigida provides one line of evidence for testing the potential benefits of body restraint and envenomation. The nearly universal use of both envenomation and coiling with hard-shell crayfish is consistent with these behaviors being elicited by attributes of prey that make them more difficult to handle. However, unexpectedly, the global effect of envenomation was to increase rather than decrease total handling time because the envenomation was often a sizeable fraction of total handling time (mean = 37%; range 6–84%). Swallowing time, which excludes envenomation time, increases significantly with increased RPA and for hard- versus soft-shell crayfish (Figure 4B), but we could not detect a significant benefit (negative effect) of envenomation on swallowing time within L. rigida. Within the soft-shell crayfish, envenomation was slightly more likely to occur with increased values of RPA and hardness, and it was associated with the occurrence of post-capture tail flips. Thus, despite increased size, hardness, and movement eliciting envenomation, the benefits of envenomation for handling prey are subtle at best based only on variation within our study species.

Comparing Liodytes rigida and its sister species L. alleni provides additional insights into the potential benefits of envenomation because they both eat crayfish and use body restraint, but only L. rigida uses envenomation (Franz, 1977; Tumlison and Roberts, 2018; Gripshover and Jayne, 2021). Unexpectedly, compared to L. alleni, L. rigida did not have a lower incidence of post-capture escapes, and the probabilities of post-capture pinching occurring and the total handling times were quite similar for a given size and type of the crayfish (Figure 4A and Table 1). Unlike L. alleni, L. rigida almost releases the crayfish while repositioning its jaws to envenomate the crayfish, and the teeth of L. rigida lack the highly specialized shape in L. alleni that is believed to facilitate holding hard prey (Rossman, 1963). Perhaps, both of these factors contribute to a slightly lower chance of crayfish escaping L. alleni. However, faster swallowing times of L. rigida compared to L. alleni (Figure 4B and Supplementary Table 5) provide some strong evidence of at least one benefit of envenomation by L. rigida. Interestingly, the processes of subjugation and consumption (Endler, 1991) are decoupled when L. rigida (or any other predator) kills its prey, whereas these processes are inseparably coupled in L. alleni and a host of other snake species that swallow their prey live. Perhaps envenomation by L. rigida also reduces the chance of internal injury, but we lack the data needed to address this.

The evolutionary origin and benefits of L. rigida using its body to restrain prey become more apparent with comparisons of other natricine species. Body restraint of prey is quite rare in the clade of North American natricines (Rossman et al., 1996; Gibbons and Dorcas, 2004), which includes three major lineages with the following genera: (1) Liodytes and their relatives, (2) Thamnophis, and (3) Nerodia and Regina (Figueroa et al., 2016). The widespread lack of body restraint implies this was the ancestral condition for this entire clade and was retained in L. pygaea. Hence, body restraint probably evolved in the common ancestor of L. rigida and L. alleni and was closely associated with another derived trait of a specialized diet of hard-shell crayfish (Figure 5). Regina septemvittata eats only soft-shelled crayfish, and they do so without any body restraint although they often hold their prey with their jaws for a substantial time before swallowing commences (Gripshover and Jayne, 2021). For soft-shell crayfish, the faster handling times of both L. rigida and L. alleni compared to R. septemvittata (Figure 4A) provide strong evidence for a benefit of body restraint for feeding performance on a given type and size of prey.

Similar to the rarity of body restraint within North American natricines, well documented cases of envenomation of prey by these snakes are also quite rare (Rossman et al., 1996; Gibbons and Dorcas, 2004) although several species of old world natricines commonly use envenomation (Weinstein et al., 2011). The absence of envenomation in both L. pygaea and L. alleni suggests that rather than being retained from a common ancestor of Liodytes, the capacity for envenomation evolved in L. rigida (Figure 5).

Unlike the independent evolutionary origins of a specialized crustacean diet within the North American natricines, all three of the genera of Southeast Asian homalopsid snakes that specialize on crustaceans form a monophyletic group (Figueroa et al., 2016). All three of these genera have retained the opisthoglyphous venom apparatus (Jayne et al., 2018) that occurs in most homalopsid species and is often used to kill vertebrate prey (Jayne et al., 1988; Murphy, 2007). Unlike L. rigida and rather paradoxically, all three of the crustacean-eating homalopsids have sizeable, grooved rear fangs, but they do not appear to envenomate their prey. Unlike crayfish, the crabs eaten by homalopsids have both ventral and dorsal surfaces that are heavily armored and likely to prevent penetration by the teeth of the snakes. Presumably teeth could penetrate soft-shell crabs, but neither Fordonia leucobalia nor Gerarda prevostiana appear to envenomate these prey (Jayne et al., 2018). Cantoria violacea eats snapping shrimp, which have a soft underside similar to crayfish, but envenomation appears absent for these prey which are usually swallowed dorsal side up (Jayne et al., 2018). The venoms of snakes can have considerable specificity as indicated by some that are highly effective for reptiles but not mammals (Mackessy et al., 2006) or more toxic for arthropods without increased toxicity for vertebrates (Starkov et al., 2007). Thus, an interesting issue to resolve in the crustacean-eating snakes is the importance of venom composition versus a mechanical barrier in prey for impeding successful envenomation.

In common with L. rigida and L. alleni, all three of the crustacean-eating homalopsid species use some form of body restraint (Jayne et al., 2018). Cantoria violacea occasionally uses its ventral surface to pin snapping shrimp. Fordonia leucobalia strikes both hard- and soft-shell crabs with a closed mouth and then uses its chin to pin crabs before quickly coiling around them. Hence, the use of body restraint in F. leucobalia is more stereotyped than L. rigida as the latter species commonly lacked body restraint for soft-shelled prey. Gerarda prevostiana only eats soft-shell crabs, and with increased prey size it relies increasingly on coiling to restrain prey to facilitate ripping them apart, which is unlike the generalized prey handling behaviors of the two species of Regina that feed exclusively on soft-shell crayfish. None of these three species that specialize on soft-shell crustaceans have any obvious morphological specializations, whereas two of the species specializing on hard-shell crustaceans (L. alleni and F. leucobalia) have very unusual, specialized dentition that appears better suited than generalized tooth morphology for coping with hard surfaces (Rossman, 1963; Savitzky, 1983). Furthermore, all four of the species specializing on hard-shell crustaceans (L. rigida, L. alleni, C. violacea, and F. leucobalia) use at least one specialized behavior for handling prey. Collectively, this system reinforces the pivotal role that behavioral innovations can have for exploiting novel, formidable prey even without accompanying anatomical specializations. In addition, the intuitively appealing evolutionary scenario that snakes ate less dangerous soft-shell crustaceans before more dangerous hard shell-prey is not supported either for the homalopsids or for the two species of Liodytes that eat hard-shell crustaceans.

In some lineages of snakes, especially those that kill prey by constriction rather than just coiling to restrain prey, the pattern of coiling and the use of coiling can be quite stereotyped (Greene and Burghardt, 1978). By contrast, this study and a previous study (Gripshover and Jayne, 2021) found both species of Liodytes that eat crayfish use their bodies in variable ways to restrain prey even though the occurrence of body restraint was quite stereotyped for hard-shell crayfish. Perhaps killing prey with coiling is functionally more demanding than merely restraining prey, and that could be a significant factor contributing to stereotypy in many constricting snakes.

As the speed of predatory behaviors increases, one might expect that the ability to incorporate feedback to modify the behavior would diminish. Although it is not as fast as some other vertebrate predatory behaviors, such as suction feeding of fish or tongue projection by plethodontid salamanders, the duration of many snake strikes still appears short enough to afford little opportunity to incorporate feedback to modify the strike after it is initiated (Cundall et al., 2007). However, pre-strike information regarding the location, type, and size of prey could allow snakes to use different preset motor programs that would not otherwise be subject to feedback (Cundall et al., 2007). The refusal of L. rigida to attack very large prey (even when they were smaller than maximal gape) and the high accuracy of strike location suggest visual cues are important for such pre-attack modulation of behavior. The high percentages of initial strikes that were associated with prey movement further support the use of visual cues by L. rigida, and visual cues are known to be sufficient to elicit attacks in several other species of natricine snakes (Drummond, 1985). Prey moving in water also generate vibrations, and vibrations elicit attacks in at least some species of aquatic snakes (Jayne et al., 1988; Catania et al., 2010). Hence, the common defensive behavior of remaining immobile (Nishiumi and Mori, 2015; Roelofs, 2017) could facilitate crayfish evading detection by L. rigida, and we observed several trials in which this appeared to be happening. Since some natricines can use olfaction to detect freshly molted crayfish (Jackrel and Reinert, 2011), L. rigida could also know the molt condition of prey prior to attack. Olfactory cues and vibrations seem likely to have additive effects with visual cues for increasing the probability of attacks by snakes on aquatic prey (Drummond, 1985). Once L. rigida contacts prey, a variety of tactile cues from the variable hardness and the vigor of prey movements could provide additional feedback for modifying the behaviors of coiling and envenomation, both of which are much slower than the strike.

The benefits of highly toxic venoms for subduing prey seem obvious for the elapid and viperid snakes with hollow fangs at the front of their mouth through which venom can be forcibly (Mackessy and Saviola, 2016) and rapidly (<200 ms) injected (Cundall, 2009). However, much remains to be understood regarding the benefits of a more rudimentary venom apparatus as occurs in L. rigida and a wide-variety of rear-fanged snakes (Weinstein et al., 2010, 2011; Mackessy and Saviola, 2016). The morphology of front-fanged snakes could permit modulating the amount of venom delivered within a fraction of a second (Cundall, 2009), whereas for snakes such as L. rigida without hollow fangs, the primary mechanism for modulating the amount of venom delivered seems likely to be how long they bite and hold their prey, which can take several minutes. Interestingly, many of these snakes lacking hollow fangs also use their body to restrain or constrict prey (Rochelle and Kardong, 1993); hence, they have some redundancy for how prey are subdued. An interesting issue for these snakes is how commonly the evolution of body restraint may have preceded venom apparatus or enhances the utility of rudimentary venom apparatus by securing prey during the prolonged process of venom delivery.

In part because of their medical importance, envenomation by viperids and elapids is better studied compared to rear-fanged snakes (Weinstein et al., 2011). However, the diverse diets, variable toxicity and specificity of venom, frequent use of body restraint and constriction, and the phylogenetic diversity of rear-fanged snakes provide a rich, model system for future study of the evolution of specialized morphologies and behaviors involved in predator prey interactions. To a great extent L. rigida used envenomation and body restraint as needed resembling many species of venomous snakes that apparently use their venom frugally (Morgenstern and King, 2013), which would require some sensory information or feedback regarding prey. However, an alternative strategy could be to be minimally reliant on feedback and use “overkill” to maximize the speed of subduing prey and minimize the risk to the predator. Testing between these two alternatives or for a continuum of variation between them may be another fruitful area for future comparative studies, especially in light of the tremendous variability in the times for which venom apparatus and specialized predatory behaviors have evolved in different lineages of snakes.

The original contributions presented in this study are included in the article/Supplementary material, further inquiries can be directed to the corresponding author.

This animal study was reviewed and approved by the Institutional Animal Care Use Committee of the University of Cincinnati (protocol number: 07-01-08-01).

Both authors contributed to the all aspects of the conceptualization, methodology, investigation, data analysis, and writing and article and approved the submitted version.